Global gas turbine orders hit a record 38 GW in Q2 2026, roughly 52% above the 25 GW quarterly average implied by 2025's 100.3 GW full-year total of 846 units, with the United States absorbing about half of the volume [S4].
Aggregating the three Western OEMs' headline backlogs, GE Vernova at 116 GW, Siemens Energy at 69 GW, and Mitsubishi Power at 35 GW, produces a combined book of about 220 GW; but only 53 GW of GE Vernova's figure is firm equipment orders, with the remaining 63 GW held as paid slot reservations that have not yet converted to contracted steel [S3][S4]. For related process equipment context, see the broader turbine flowmeter reference set, since metering sits on the same combustion-package procurement path as the prime mover.
Where the 220 GW actually sits: firm orders vs paid options
GE Vernova's Power segment booked $16.7 billion in orders during Q2 2026, up 134% organically, lifting gas equipment backlog plus slot reservations to 116 GW, with 52 heavy-duty units (15 of them HA-class) and 61 aeroderivative units signed during the quarter [S3]. The company also converted 10 GW of pre-existing slot reservations into firm orders and shipped 3 GW of equipment, leaving 53 GW of firm backlog and 63 GW of reservation slots at quarter-end [S3].
Mitsubishi's restructured Energy Systems division saw order intake rise 56% year over year in Q1 of its fiscal 2026, booking 10 large-frame orders in the quarter (four for the United States, six for Japan) and lifting its large-frame backlog to 35 GW from 23 GW a year earlier; Mitsubishi sold 35 large-frame units totaling 16 GW in fiscal 2025 [S1]. The new units are scheduled for delivery between 2028 and 2030, and Mitsubishi still plans to double its 2024 production capacity, though CFO Hiroshi Nishio said the company is "being selective" on which projects it contracts [S1]. Siemens Energy's reported 69 GW, as cited in the SecondWatt industry write-up, places it second by headline number but uses the same firm-plus-reservation accounting convention as GE Vernova [S4].
The factory ceiling: 60–70 GW/yr against 110–125 GW of orders
Wood Mackenzie pegs global end-2025 orders at 110 GW against worldwide manufacturing capacity of 60–70 GW per year, which means the existing book absorbs 1.6–1.8 years of full-rate production before a new order gets touched [S4]. GE Vernova's own expansion plan targets 20 GW of annualized gas turbine output in Q3 2026, 24 GW in 2028, and 30 GW in 2030, with manufacturing expansion being executed largely within the existing footprint [S3]. Mitsubishi has separately committed to doubling its 2024 capacity, with detailed growth-plan language deferred to the second half of fiscal 2026 [S1].
Year-over-year capacity additions are accelerating from 8.5 GW actually realized in 2025 to 13.6 GW scheduled for 2026 and 36.3 GW scheduled for 2027, with hot-section castings repeatedly identified as the chokepoint below the factory-level ceiling [S4][S8]. U.S. developers alone plan to add 18.7 GW of combined-cycle capacity to the grid by 2028, of which 4.3 GW is already under construction, per the U.S. Energy Information Administration's June update [S5].
Lead times by frame class: 5 years large, 18–36 months small
Large-frame gas turbines are now quoting around five years from order to commissioning, while small-frame units reportedly run 18–36 months, a split that makes frame class the single most important procurement-strategy variable for any project with a commercial operation date before 2029 [S4]. NextEra Energy Chairman, President and CEO John Ketchum said at a June 2025 Politico Energy Summit that a new gas-fired facility "can't get it online until 2032," and that gas turbine prices have risen sharply in the prior 24 months [S5].
Wood Mackenzie tracks turbine prices up 195% since 2019 and heading toward roughly $600/kW by end-2027; turbines typically represent 20–30% of a combined-cycle project's cost and a larger share of a simple-cycle project's cost [S4]. On a per-unit basis, the Q2 2026 surge at GE Vernova worked out to an average of about $148 million per combined heavy-duty and aeroderivative order signed, weighted by class mix; for comparison, the company's Q4 2025 cohort of 41 heavy-duty plus 18 aeroderivative units was valued at $7.7 billion, or about $130 million per unit [S3][S4].
Who is buying: AI data centers, grid reliability, and a 4.4 GW anchor
Data center orders alone topped $5 billion year-to-date at GE Vernova by end of Q2 2026, more than double the full-year 2025 figure, while the broader Electrification segment rose 66% organically to $6.3 billion on grid equipment demand [S3]. Q1 2026 specifically logged $2.4 billion in AI data-center gas turbine orders, and one analyst write-up pegged revenue potential per GW at 2–3x the historical baseline as solid-state transformers and medium-voltage UPS blocks approach commercialization [S3][S7].
On the buyer side, Homer City Redevelopment announced an agreement in principle with EQT Corporation to supply gas to a 4.4 GW natural gas facility serving the Homer City Energy Campus, a 3,200-acre AI and high-performance computing site in Pennsylvania slated to begin producing power in 2027 [S5]. Frontier Group of Companies is converting the former 2.7 GW Bruce Mansfield Power Plant into a larger natural gas plant, Shippingport Power Station, with a collocated data center [S5]. These U.S. anchor projects sit alongside large orders from Saudi Arabia, Asia, and Western Europe tied to LNG infrastructure, all of which Ketchum cited as competing for the same factory slots [S5]. For ancillary package engineering tied to these builds, the construction machinery and equipment index covers the erection fleet these projects depend on.
How the three OEMs compare on backlog, capacity, and disclosure
On a like-for-like headline basis, GE Vernova leads at 116 GW, Siemens Energy follows at 69 GW, and Mitsubishi Power sits at 35 GW, but the disclosure conventions differ: GE Vernova breaks out 53 GW firm versus 63 GW reservation, while Mitsubishi reports a unified 35 GW large-frame figure and a 56% year-over-year order-intake jump [S1][S3][S4]. On capacity expansion, GE Vernova has the most quantified ramp, with explicit 20 GW / 24 GW / 30 GW annual-output targets for 2026 / 2028 / 2030 and a footprint-locked strategy using about 400 new factory machines by year-end 2026, while Mitsubishi has announced a doubling of 2024 capacity without yet publishing a calendar-year output target [S1][S3]. On lead-time exposure, all three OEMs are sold out for 2026 and 2027 delivery in the large-frame class, with Mitsubishi's first-quarter orders explicitly scheduled for 2028–2030 delivery [S1][S5].
Limitations of the headline number: reservations, cancellations, and the EPC labor squeeze
Add GE Vernova's 116 GW, Siemens Energy's 69 GW, and Mitsubishi's 35 GW and you get 220 GW, but that figure is not directly comparable across vendors because none of the three is counting the same thing: GE Vernova's 63 GW reservation slice and the equivalent portions of the other two books are paid options, not contracted equipment [S4]. A further constraint sits below the factory: Ketchum noted it takes about a thousand people to build a gas-fired generation facility, that EPC contractors must hire roughly six thousand to net a thousand, and that washout is driven by competing data center, semiconductor, LNG, and chemical construction demand [S5]. For instrumentation and gas-path measurement scope on these builds, see the checking and testing apparatus reference cluster covering the QA chain for combustion packages. For emission monitoring on the same plants, the gas analyzer page covers CEMS and process-gas analysis hardware typically specified alongside new turbines.
Trackable signals over the next two quarters: GE Vernova's stated 125 GW under contract by year-end 2026 versus its current 116 GW; Mitsubishi's second-half growth-plan update with concrete output tonnage; and EIA combined-cycle capacity-addition revisions in the 18.7 GW by 2028 figure. Any one of those moving materially will reset the 1.6–1.8 years-of-supply figure that the procurement market is currently planning against [S3][S4][S5].
For related coverage, see ACI 506R Shotcrete Aggregate Gradation: Limits, Air Content, and Specification Choices.